Flood Forecasting February 11th, 2015 A Statistical Analysis and Geospatial Distribution of Soil Moisture in Texas Gonzalo E. Espinoza Center for Research in Water Resources Flood Forecasting February 11th, 2015
Outline Land Assimilation System (LDAS) Statistical Analysis NLDAS NOAH and soil moisture Statistical Analysis Cumulative Distribution Functions (CFDs) Applications Texas Drought 2011 “Halloween Flood”. Onion Creek, 2013 Web Application Architecture Client Server Cloud Work in progress
Land Assimilation System (LDAS) Land-surface hydrologic model Global (GLDAS ) National (NLDAS) NOAH NLDAS-2 Variables ~ 52 Hourly and Monthly data 1/8° cellsize ~14km Data since 1979 http://ldas.gsfc.nasa.gov/nldas/NLDASgoals.php
NLDAS NOAH & Soil moisture Soil Moisture in the top meter Top meter (4 layers) 0-10cm 10-40cm 40-60cm 60cm-100 Kg of water/m^2 Equivalent depth of water (mm) http://ldas.gsfc.nasa.gov/nldas/NLDAS2model.php
Empirical CFD calculation Statistical Analysis Soil Moisture -Variable -Time -Space Time Series Quad: 30097-C6 Date: 12/04/2014 NLDAS Data CDF modelling Empirical CFD calculation CDFs construction -Time step (day) -Cell size (1/8°) -Soil Moisture -Evapotranspiration -Precipitation -Runoff -Temperature Cell Day Variables Date: June, 15th Cloud Storage Validation Implementation Python/R HPC Results Cloud Web app http://texassoilmoisture.azurewebsites.net
Cumulative Distribution Functions (CFDs) Daily values for each cell 4,136 cells 1979 – 2013 Cumulative Distribution Functions (CDFs) For each cell and each day Statistical parameters (min, max, mean, standard deviation, etc.) Comparing the latest soil moisture value Cell Day of the year Examples: West: El Paso Central: North of Austin Coast: South of Houston
Cumulative Distribution Functions (CFDs) Coast: South of Houston West: El Paso Central: North of Austin
Cumulative Distribution Functions (CFDs) West Central Coast
Applications: Texas Drought 2011 September 2011 5% percentile Anomaly -55mm
Applications: Onion Creek, 10/31/2013 Drainage Area: 730km2 Pre-storm soil moisture 87% percentile
Web Application Architecture Azure cloud Storage Django ArcGIS Mapping Javascript NASA Data rods Microsoft Azure Storage Account Django (python) ArcGIS server ArcGIS API for JavaScript Client-Side Cloud deployment Server-Side User Web App Hydrologic Data NLDAS web services Geographic data Latest Values Site & Date stats Web Deployment Web app framework Stats http://texassoilmoisture.azurewebsites.net
Work in progress Process more variables Update the website design Precipitation Runoff Evapotranspiration Temperature Update the website design Layout Latest values Data storage communication Improve the performance of the cloud storage Risk scenarios Relationship between variables